2. Legal Framing Innovation
2.2 Explaining Legal Framing Innovation
2.2.4 Internal Movement Characteristics
28
hypothesize that legal framing innovation is more likely to occur in legal contexts that are more hostile towards feminists.
Prior research strongly suggests that the perspectives of individual justices influence the trajectory of legal mobilization, and activists, usually aware of judges’ ideologies, plan their legal framing strategies accordingly (Andersen 2004; McCammon & Beeson-Lynch, 2020).
There is a great deal of research that indicates that judicial decisions/trial outcomes tend to reflect ideological preferences of judges (Segal & Cover 1989; Segal 1997; Segal & Spaeth 1993; Segal, Epstein, Cameron, & Spaeth 1995; Gely & Spiller 1990; Spiller & Gely 1992;
Epstein 1995; de Figueiredo & Tiller 1996; McCubbins, Noll, & Weingast 1989, 1995; Cross &
Tiller 1998), and studies commonly use the Martin-Quinn score as a measure of judicial ideology.
29
definitions to invoke claims on behalf of a movement’s members (Snow & Benford 1988). To that end, scholars have empirically demonstrated that movement and organizational identity26 influence framing strategies (Evans 1997; Steinberg 1999). For example, Rohlinger (2002) found that while NOW altered its framing strategies in response to changes in the broader political and cultural context, antichoice SMO, Concerned Women for America (CWA) did not. Her
interviews with SMO representatives reveal that organizational identity likely plays a role in explaining the differences in framing strategies. According to a CWA representative, its members viewed abortion as a moral absolute that did not change over time (Rohlinger 2002:492).
Scholars that study interest group litigation have examined a range of organizational factors including the available legal organizational resources and group ideology. In terms of the former, there is a great deal of research that explores the role of resources as an important factor in shaping a group’s litigation strategy and decision-making (Wasby 1995; Songer, Kuersten &
Kaheny 2000; Epp 1998)27. In regard to group ideology, several researchers empirically demonstrate that organizational identity and group ideology shape litigation strategy (Vanhala 2009; Levitsky 2007; Krishnan & den Dulk 2002). Both areas of research have important implications for the effects of internal movement characteristics on legal framing innovation in abortion clinic protest cases. The prochoice movement has a long history of engaging in legal advocacy, and their experience may enable them to absorb immediate losses relatively more easily than their opponents. This experience may also help them gather intelligence and establish
26 McCammon (2012:48) defines organizational identity as a “group’s core political values and assumptions.”
27 External groups influence judicial decisions, but “only where the support structure is sufficiently strong to generate continued litigation... [because it can] reassure judges that if they do support rights claims, they will not be bereft of allies in the event of political attack” (Epp 1998:201).
30
long-term alliances with other organizations (see Galanter 1974; McCann 1994; Handler 1978).
On the other hand, some of the most prominent religious groups in the abortion protest coalition have only recently begun to engage in legal advocacy (Krishnan & den Dulk 2002; Hoover &
den Dulk 2004; Heinz et al. 2003). Moreover, movement ideology provides its supporters with a framework for how to identify, understand, and combat a particular issue or set of issues by linking a movement’s identity with the frames that its members adopt (Hunt, Benford, & Snow 1994; Klandermans 1997). By their very definition, framing processes involve the use of shared meanings and definitions to invoke claims on behalf of a movement’s members (Snow &
Benford 1988). To that end, scholars have empirically demonstrated that movement and organizational identity28 influence framing strategies (Evans 1997; Steinberg 1999; Esacove 2004).
For example, Rohlinger (2002) found in her interview data with Concerned Women for American (CWA) representatives that organizational identity likely plays a role in explaining framing strategies. According to a CWA representative, its members viewed abortion as a moral absolute that did not change over time and the group then consistently framed the issue as the murder of unborn babies, despite changes in the political context and its opponents tactics (Rohlinger 2002). On the other hand, Rohlinger (2002) found that the National Organization for Women (NOW), a public policy group that supports abortion access, typically framed abortion in terms of protecting women’s rights which was in line its organizational identity. As such, I
28 McCammon (2012:48) defines organizational identity as a “group’s core political values and assumptions.”
31
hypothesize that legal framing innovation is more likely to occur in feminist- and opponent- supporting briefs when there is an increased presence of SMOs committed to a specific cause29.
29 For example, if there was an increase in the proportion of SMOs whose broader ideological- and organizational-agendas include a commitment to protecting the lives of the unborn, then I would expect an increase in the use of legal frames that emphasize the need to restrict abortion in order to protect fetal life.
32 CHAPTER 3
Data and Methods
In this chapter I discuss the data and variables that I use to study legal framing in amicus and party briefs associated with nine U.S. Supreme Court cases related to abortion clinic
protests. The nine abortion protest cases I consider are Frisby v. Schultz (1988), Bray v.
Alexandria Women's Health Clinic (1993), NOW v. Scheidler (1994), Madsen v. Women's Health Center (1994), Schenck v. Pro-Choice Network (1997), Hill v. Colorado (2000), Scheidler v.
NOW (2003), Scheidler v. NOW (2006), and McCullen v. Coakley (2014). I begin this chapter by discussing the data I use to construct the dependent variable in my regression analysis (Chapter six). Then I discuss the data and variables that I use to examine legal framing variation in Chapters five and six of my dissertation.
3.1 Dependent Variable: Framing Innovation
In Chapters four, five, and six, I use text mining, machine learning, and multilevel logistic regression to study legal framing in amicus and party briefs submitted in nine Supreme Court cases related to abortion clinic protests. The amicus and party brief data come from a larger project that I worked on under the direction of Professor Holly McCammon and with research assistants in addition to myself. The data were gathered using multiple sources:
LexisNexis Academic/Nexis Uni, Westlaw, and United States Reports30. I situate the framing activity in the broader political, cultural, and legal contexts in Chapters five and six to examine how the external environment influences legal framing strategy. In the regression portion of my
30 The last of these is the official case reports of Court decisions.
33
analysis (Chapter six), I also consider how internal movement characteristics affect legal framing.
The legal brief data include a total of 182, party briefs, including reply- and
supplemental-briefs, submitted by the parties, and amicus briefs filed in the nine abortion clinic protest cases. Party briefs are written documents in which the attorneys present their arguments to the Court. Party reply briefs provide litigants with an opportunity to refute their opponents’
arguments and persuade the Court to decide in their favor (Vail 2000; Wolfman 2023). Parties to the case may also file a supplemental brief that calls attention to new cases or other relevant matter while a petition for the Court to hear the case (i.e., writ of certiorari) is pending31. Amicus curiae, or “friend of the court,” briefs are briefs submitted to the Court by individuals and/or groups that are not directly involved with the legal case (Collins 2018). These briefs allow amici to present new or different legal positions and perspectives to the Court (Banner 2003; Harris 2000).
Figure 1 illustrates the distribution of the number of briefs by case (in chronological order, by the Supreme Court decision date), brief type (party, party reply and supplemental arguments, or amicus brief), and supporting party (either feminist or opponent). Of the 71 legal briefs that support the feminists in the cases, nine are party briefs, one is a party reply brief, and 61 are amicus briefs. There are 111 briefs that support the opponents (i.e., abortion clinic protesters), and of those 111, 12 are party briefs32, 15 of the briefs are party replies and
supplemental arguments, and 84 are amicus briefs. Over time, the total number of feminist- and
31 Per U.S. Supreme Court Rule 15.
32 In Scheidler v. NOW (2003) and Scheidler v. NOW (2006), the petitioners, Joseph Scheidler and the Pro- Life Action League (PLAN), wrote separate party briefs.
34
opponent-supporting amicus briefs has increased, and in most of the cases, the total number of opponent briefs, including both party and amicus briefs, outnumbers the feminist briefs.
Figure 1 Number of legal briefs by brief type, supporting party, and legal case
3.1.1 Protect Life Legal Frame
In order to examine legal framing, I focus on a how parties on both sides of the cases emphasize the need to protect the lives of particular groups of citizens, or what I refer to as
“protect life” legal frames. There are two reasons why I focus on the “protect life” frame. First, during my close reading of the briefs, I found that parties and amici on both sides of the cases often referred to the state’s compelling interest in protecting the life of one or more groups in society to further their positions in the abortion clinic protest cases. This is because the cases in
35
my study deal with questions about First Amendment and Fourteenth protections33 available to protesters at health clinics providing abortion services34. The U.S. Supreme Court has a
longstanding tradition of protecting citizen’s right to free speech under the First Amendment (Gora 2016). For the government to curb expression, it must convince the Court that restricting speech is the least drastic means of achieving its compelling interest in eliminating the threat of serious harm (e.g., the threat to public safety). In order to accomplish this, the government must persuade the Court of each: (1) the compelling nature of the government’s interest; (2) the absence of a less restrictive means to protect its interest; (3) the seriousness of the harm; (4) the inevitability of the harm; (5) a clear causal link between the speech and the harm; (6) the restriction’s effectiveness in achieving the government’s interest in eliminating the harm.
The second reason I chose to focus my attention on the “protect life” frame is because I noticed that framing innovation appears in briefs supporting both sides of the cases when this argument is invoked, and in Chapters four and five, I use text mining and machine learning to identify instances of “protect life” framing innovation.
In order to identify legal framing innovation in the nine Supreme Court cases in my study, I began with close and multiple readings of all party and amicus briefs to generate a list of words and phrases associated with the “protect life” frame in Atlas.ti. My coding scheme was iteratively developed in several stages using inductive reasoning and open coding at the sentence level (Ferree et al. 2002; Nelson et al. 2018).
33 Bray v. Alexandria Women’s Health Clinic (1993) does not deal with the First Amendment, but rather it concerns questions regarding gender discrimination and the application of the Fourteenth Amendment (see Banks 1994; Campbell 1993; Fischer 1993).
34 Frisby v. Shultz is the only exception to this, and it deals with protests outside the home of a local doctor who performed abortions.
36
I chose to analyze the legal briefs at the sentence level for three reasons. First, the machine learning model that I use to identify legal frames in Chapter five can only handle texts that are less than or equal to 512 characters, and the average length of a legal brief in my dataset is 38,019 characters whereas the average sentence length is 155 characters. Second, classifying sentences rather than briefs increased the size of my text classification dataset from 182 legal briefs to 1,502 sentences in the text mining (Chapter four) and machine learning (Chapter five) portions of my analyses, and it is well-documented that enlarging the sample size can help reduce over-fitting (see Turton et al. 2020 and Zhao et al. 2022).
My coding scheme also allowed me to distinguish between the different ways that
litigants use the “protect life” frame. During my close reading and open coding of the legal briefs in Atlas.ti, I noticed that the brief authors pointed to several distinct groups that needed
protection. In order to capture these differences in Atlas.ti, I categorized sentences based on whom the sentence claimed needed protection. After several iterations of this process, I found that most sentences fell into one or more of the following five categories: protecting the life of (1) women; (2) unborn; (3) clinic workers; (4) patients; and (5) public. If a sentence did not receive any of these codes, it received the code, “no frame.” To receive any of the “protect life”
codes, a sentence must explicitly discuss the need to defend the physical health and/or safety of one of the five groups. That is, if a sentence discusses protecting rights or access to clinics then it would not receive a “protect life” code. Each sentence could receive either a single, “no frame,”
code or up to five protecting life codes, although in practice, the maximum number of codes a sentence received was three.
In order to construct this measure, I manually coded 1,105 (2.5%) of the 44,711 sentences in my legal brief dataset (Chapter four), and I used a masked language model, a process that I
37
discuss in more detail in Chapter five, to classify the remaining legal brief sentences. After classifying the sentences, I built a multilevel regression model in order to study the impact of the legal, political, cultural, and internal movement contexts on the use of the “protect life” frame.
In Chapter six, I use multilevel logistic regression to predict legal framing innovation, and I use the appearance of the “protect the lives of patients” frame as the dependent variable in my model. The binary measure is equal to 1 if the frame is the most common “protect life” frame (i.e., primary frame) in the brief, and 0 otherwise. As such, the unit of analysis in Chapter six is the legal brief, and there are 182 amicus and party briefs filed in the nine abortion clinic protest cases.
I chose to use incidents of the “protect the lives of patients” frame as a measure of legal framing innovation in my regression analyses because it allows me to explore what might cause a particular legal frame to be used. That is, by predicting if the “protect the lives of patients” will appear as the most common “protect life” frame in a brief, I can empirically study which
external- and internal-movement factors constrain or facilitate instances of a particular legal frame.
However, while this measure can aid in understanding what contributes to use of the
“protect the lives of patients” frame, this measure does not allow us to see the first use of the frame. That is, because this is a cross-sectional measure, a measure that indicates use of the frame in different legal briefs, it does not tell us what causes the frame to be used for the first time35. Thus, nor can this measure tell us what causes variation in movement framing strategy.
As such, it leaves open the question of what causes legal framing innovation to occur. As several
35 To explore influences on changes over time (i.e., framing innovation), future work could consider building a longitudinal measure.
38
researchers have noted (Johnston 1995; Steinberg 1998), it is still unclear how “frames get made” (Hart 1996:95); I argue that my dependent variable and the results in Chapter six help to address this gap in the literature by identifying important internal- and external-movement factors that influence movement framing strategy.
3.2 Independent Variables and Measures
In this section, I describe the data and variables that I use to situate the findings of the machine learning model in Chapter five. These are also the variables that I use in my regression analysis (Chapter six) to examine how political, cultural, legal, and internal movement contexts may influence legal framing in amicus and party briefs in all nine US Supreme Court abortion clinic protest cases. Table 1 summarizes the variables (Column two) according to their context (Column one). In addition to the broader contextual- and internal-movement factors, I also include the brief’s supporting party (feminist or opponent) and brief type (party or amicus) as independent measures in the multilevel logistic regression models in Chapter six.
39
Context Variable
Political Proportion of Democrats in Senate Proportion of Democrats in House
Proportion of states with Democrat majority in both state-level chambers Proportion of states with Democrat majority in at least one state-level chamber Cultural Average sentiment of conservative newspapers
Average sentiment of liberal newspapers
Average sentiment of abortion newspaper articles
Average sentiment of abortion protest newspaper articles Legal Average Martin-Quinn score
Direction of lower court ruling (1 = favors feminists; 0 = favors protesters) Case type (1 = buffer-zone case; 0 = not buffer-zone case)
Internal Proportion of medical authors
Proportion of right social movement authors Proportion of left social movement authors Proportion of government authors
Proportion of legal authors Proportion of religious authors Proportion of labor union authors
Other Supporting party (0 = opponents, 1 = feminists) Brief type (0 = party brief36; 1 = amicus brief)
Table 1 Independent variables
36 This includes party briefs, party reply briefs, and supplemental argument briefs submitted by parties on either side of the case.
40 3.2.1 Political Context
I hypothesize that legal framing innovation is more likely to occur in conservative political contexts that are more hostile to feminists when controlling for the brief’s supporting party and type (either amicus or party). In order to test this hypothesis, I include four measures to explore the role of the political context in influencing legal framing. The first two are measures of the percent of Democrats in the House of Representatives and the Senate in the year that the Court decides the case37. The third measure is the proportion of states with a Democrat majority in both state-level congressional chambers, and the fourth measure is the proportion of states38 with a Democrat majority in at least one state-level congressional chamber.
Figure 2 illustrates the political context measures over time by legal case.
Broadly, the political context has become more conservative and, as such, become more hostile towards pro-choice legal arguments. Over time, the Democrats have lost majority control over the House and Senate, although they did regain control of the Senate when the Court heard McCullen v. Coakley (2014). However, the proportion of states with a Democrat majority in at least one state-level chamber has decreased over time; again, with a slight increase during the time period between when the Court decided Scheidler v.
NOW (2006) and McCullen.
37 The House and Senate both maintain public statistics of party divisions immediately following national elections (House: https://history.house.gov/Institution/Party-Divisions/Party-Divisions/ and Senate:
https://www.senate.gov/history/partydiv.htm)
38 Nebraska is omitted because it is a non-partisan, unicameral legislature.
41
Figure 2 Political context: Tracing measures by legal case
3.2.2 Cultural Context
Public support for abortion has remained relatively stable since the Court ruled on Roe v.
Wade (1973), and Democrats and those who lean toward the Democratic Party tend to think that abortion should be legal and accessible to everyone. On the other hand, Republicans and those who lean toward the Republican Party tend to favor policies that restrict people’s access to abortion (Pew Research Center 2022); often these restrictions are upheld on the basis of protecting women’s health39.
Therefore, I hypothesize that feminist and opponent lawyers are more likely to put forth legal arguments that emphasize the health and safety of patients when conservative attitudes toward abortion and abortion protests are more negative. Conversely, when liberal attitudes toward abortion and abortion protests are more positive, I expect lawyers on both sides of the
39 See Friedman (2013:48) for further discussion on how the Supreme Court’s jurisprudence since Roe v.
Wade (1973) has facilitated the “abuse of the argument that the law is protecting women’s health as a pretense to limit access to abortion.”